Uploaded March 2026 | Updated September 2026, 10 hours ago

---

MCP is one of the best ways to extend AI agents and chatbots across your organization — but your MCP servers still need authentication and access control.

In this tutorial, we walk through securing an MCP server using WorkOS AuthKit.

You’ll learn how to add authentication to a FastMCP server, configure OAuth, protect MCP tools with scopes, and test authentication end-to-end.

By the end of this video, you’ll have a secure MCP server connected to WorkOS AuthKit and ready to integrate with your AI agents.

Chapters:

00:00 Why MCP Servers Need Authentication
00:26 Add WorkOS AuthKit to a FastMCP Server
01:21 Configure OAuth and Environment Variables
02:34 Authenticate and Test Your MCP Tool

⸻

What you’ll learn
• How MCP servers work with AI agents
• Adding AuthKit authentication to a FastMCP server
• Configuring AuthKit domain and client ID
• Securing MCP tools with authorization scopes
• Setting OAuth redirect callbacks
• Running and testing an authenticated MCP server
• Connecting MCP tools to an authenticated client

⸻

Resources
• WorkOS AuthKit docs
• MCP documentation
• FastMCP project

⸻

What’s next

This tutorial only scratches the surface. With WorkOS you can also:
• Store encrypted credentials using Vault
• Manage authorization across MCP tools
• Connect external services using Pipes